AMD takes over MEXT for memory tiering tech that enables flash to appear as DRAM to applications — tech to 'address growing memory constraints' in the data center

AMD
(Image credit: AMD)

AMD on Monday announced that it had acquired MEXT, a startup that developed a memory tiering technology that makes NAND flash memory appear as DRAM to the operating system, which enables operators of data centers to save money on DRAM. AMD expects the acquisition to help customers improve system efficiency, lower operating costs, and deploy large-scale workloads more quickly.

As AI models continue to expand and datasets grow larger, memory availability has become an increasingly important factor affecting overall system performance. In many cases, memory resources, not CPUs or GPUs, are becoming a performance bottleneck. Meanwhile, in many cases DRAM is used inefficiently.

MEXT addresses memory efficiency challenges with an AI-based memory tiering technology that moves infrequently accessed data from expensive DRAM to NAND storage, which costs orders of magnitude less per unit of capacity, and in a way that's transparent to applications. MEXT's Predictive Memory Engine continuously analyzes memory access patterns and uses AI models to anticipate which data stored in flash will be needed next. Those memory pages are proactively transferred back into DRAM before applications request them and enable software to access data as though it were in main memory, thus preserving performance levels.

Latest Videos From

By increasing the amount of usable memory available to applications, MEXT's technology aims to improve utilization of existing infrastructure and at the same time reduce needs for expensive DRAM. This approach can potentially lower total cost of ownership for cloud providers and enterprise customers and enable larger workloads to run on existing hardware. AMD believes that these capabilities can benefit both traditional data center applications and modern AI deployments, where access to large memory pools is often critical for efficiency and scalability.

AMD plans to incorporate MEXT's technology into its data center product portfolio and expand its capabilities to address memory-hungry AI workloads. The company already offers integrated solutions that combine processors, accelerators, networking technologies, and software, so MEXT's Predictive Memory Engine will complement the already broad portfolio.

As an added bonus to the technology itself, AMD gains a team with expertise in memory architectures, infrastructure software, and large-scale computing systems. Terms of the deal are unknown.

Google Preferred Source

Follow Tom's Hardware on Google News, or add us as a preferred source, to get our latest news, analysis, & reviews in your feeds.

TOPICS
Anton Shilov
Contributing Writer

Anton Shilov is a contributing writer at Tom’s Hardware. Over the past couple of decades, he has covered everything from CPUs and GPUs to supercomputers and from modern process technologies and latest fab tools to high-tech industry trends.

  • DS426
    So speculative execution but for memory? Speculative memory?
    Reply
  • hotaru251
    so AMD wants to increase ssd prices to "slightly" reduce amount of RAM (which totallllly is going to happen...)
    Reply
  • usertests
    hotaru251 said:
    so AMD wants to increase ssd prices to "slightly" reduce amount of RAM (which totallllly is going to happen...)
    I don't think you even read the headline, much less the article.

    By increasing the amount of usable memory available to applications, MEXT's technology aims to improve utilization of existing infrastructure and at the same time reduce needs for expensive DRAM. This approach can potentially lower total cost of ownership for cloud providers and enterprise customers and enable larger workloads to run on existing hardware.
    Reply
  • nrdwka
    ah optane manufacturing died too soon
    Reply
  • usertests
    nrdwka said:
    ah optane manufacturing died too soon
    Scary amounts of bad timing over at Intel Corporation.
    Reply
  • bit_user
    DS426 said:
    So speculative execution but for memory? Speculative memory?
    There have always been some predictive I/O optimizations, like read-ahead. It's just that I don't know how sophisticated they've gotten in the realm of virtual memory.

    The last I heard about predictive page migration was in reference to CXL, where you might have fast (i.e. high-bandwidth, low-latency) memory close to the CPU and then your expansion is sitting out over a CXL/PCIe bus. I figured that's what we'd be talking about, not predictive migration to/from swap.
    Reply
  • bit_user
    hotaru251 said:
    so AMD wants to increase ssd prices to "slightly" reduce amount of RAM (which totallllly is going to happen...)
    NAND is denser and costs less per GB than DRAM. It makes sense to use it as a lower-cost substitute for DRAM, to the extent you can.

    Also, add memory compression into the mix, and can get even more mileage out of this technology.
    Reply
  • bit_user
    nrdwka said:
    ah optane manufacturing died too soon
    Optane wasn't cost-competitive or comparable in density with NAND. The two are related, actually. That's the main reason it died.

    Even if it were still around, it'd probably just be a tier in between DRAM and NAND, but wouldn't avoid the need or the main benefits of having that NAND tier.

    Seriously, you can cram 245 TB of NAND in a single drive bay! DRAM can't touch that kind of density, neither could Optane.
    https://www.storagereview.com/review/micron-6600-ion-245tb-ssd-review-a-quarter-petabyte-per-drive-bay
    Reply
  • hotaru251
    usertests said:
    I don't think you even read the headline, much less the article.
    i did, however this also will increase ssd prices and doesnt guarantee giant infinite scaling tech giants cut back on ram as just means more growth. (and supply is whats holding them back atm)

    You forget that the megacorps dont care about sharing they will just keep gobbling up everything regardless. if they need less ram and more storage? just means they can buy more storage (and more ram) and split ram to even more servers.
    Reply
  • usertests
    hotaru251 said:
    i did, however this also will increase ssd prices and doesnt guarantee giant infinite scaling tech giants cut back on ram as just means more growth. (and supply is whats holding them back atm)

    You forget that the megacorps dont care about sharing they will just keep gobbling up everything regardless. if they need less ram and more storage? just means they can buy more storage (and more ram) and split ram to even more servers.
    More supply will come online to meet the demand, or demand will drop if the bubble deflates.

    There's been talk about High Bandwidth Flash and similar technologies for a while now so AMD is not changing the rules of the game here.
    Reply